The project involves using NVIDIA's self-driving car model and CNNs to map raw pixels from a front-facing camera to the steering commands for a self-driving car.
The project involves training the entire processing pipeline needed to steer an automobile, rather than just focusing on pattern recognition.
A small amount of training data was sufficient to train the car to operate in diverse conditions, including highways, local roads, and even unpaved roads and parking lots.
The project aims to improve the robustness of the network, find methods to verify the robustness, and improve the visualization of the network-internal processing steps.
Sample Augmented Training Image Model can drive without bumping into side-ways
